Potential adverse effects of negative publicity surrounding antivirals for influenza.

Influenza is not trivial. During the first post-pandemic winter (2010-11), there were 8797 hospital admissions for influenza in England alone. At the peak, there were 851 influenza cases in critical care beds (~20% of UK capacity), mostly younger adults and children with pandemic H1N1. Prehospital antiviral use was uncommon (<3%), yet antibacterial drugs were often prescribed (25% in our series; unpublished data). The latest Cochrane review of neuraminidase inhibitors for influenza was impressive but did not investigate antiviral effectiveness in severe influenza, illness caused by pandemic H1N1, or in patients with risk factors for complications. Therefore, the new findings cannot guide decisions about future antiviral stockpiling or the treatment of severe influenza. Those planning for a pandemic had to rely on seasonal flu data from randomised controlled trials. We now have abundant evidence from the 2009-10 pandemic, including a systematic review of observational study data from >29 000 patients admitted to hospital. After adjusting for confounding, antiviral therapy was associated with significant reductions in mortality in adults, most notably a 50% reduction with early treatment. The clear discrepancy between the Cochrane group’s conclusions and global experience of antivirals suggests that something is amiss. Observational studies are vulnerable to bias, but stringent systematic reviews of trials can mislead if the findings are not interpreted in an appropriate context. We serve patients best by looking at, and scrutinising, all available data, while understanding and accepting the limitations of different methods. By accepting only the narrowest definitions of acceptable evidence, we would deny patients effective treatments and endanger public health. The Cochrane Collaboration should engage with groups developing clinical trial protocols for future outbreaks but should also be balanced in communicating evidence. Clinical trial data must be transparent and open, but this issue should not be conflated with the deployment of safe and effective antiviral drugs.